In an Internet of Things (IoT) ecosystem, millions of devices interact to provide services in a distributed environment. However, due to the diversity and heterogeneity of IoT devices and their owners, evaluating the trustworthiness of these devices has become a significant challenge. This paper proposes a machine learning-based approach to generate trust scores for IoT service provider devices using the Support Vector Regression (SVR). The dataset includes device characteristics and the device owner’s reputation, using these features an efficient model is created to compute trust scores to assist decision-making processes. The paper also evaluates the performance of the trust prediction model in terms of MSE and \(R^2\) and of the entire framework based on accuracy, precision, and recall. Machine learning models are more suitable for IoT trust value prediction due to their efficiency in resource-constrained environments, requires less computational power and data, making them ideal for IoT devices with limited capacity.